source: branches/3044_variableScaling/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Regression/RegressionProblemData.cs @ 18242

#region License Information
/* HeuristicLab
 * Copyright (C) Heuristic and Evolutionary Algorithms Laboratory (HEAL)
 *
 * This file is part of HeuristicLab.
 *
 * HeuristicLab is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * HeuristicLab is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
 */
#endregion

using System;
using System.Collections.Generic;
using System.Linq;
using HeuristicLab.Common;
using HeuristicLab.Core;
using HeuristicLab.Data;
using HeuristicLab.Parameters;
using HEAL.Attic;

namespace HeuristicLab.Problems.DataAnalysis {
  [StorableType("EE612297-B1AF-42D2-BF21-AF9A2D42791C")]
  [Item("RegressionProblemData", "Represents an item containing all data defining a regression problem.")]
  public class RegressionProblemData : DataAnalysisProblemData, IRegressionProblemData, IStorableContent {
    protected const string TargetVariableParameterName = "TargetVariable";
    protected const string ScaleInputsParameterName = "Scale Inputs";
    public string Filename { get; set; }

    #region default data
    private static double[,] kozaF1 = new double[,] {
          {2.017885919, -1.449165046},
          {1.30060506,  -1.344523885},
          {1.147134798, -1.317989331},
          {0.877182504, -1.266142284},
          {0.852562452, -1.261020794},
          {0.431095788, -1.158793317},
          {0.112586002, -1.050908405},
          {0.04594507,  -1.021989402},
          {0.042572879, -1.020438113},
          {-0.074027291,  -0.959859562},
          {-0.109178553,  -0.938094706},
          {-0.259721109,  -0.803635355},
          {-0.272991057,  -0.387519561},
          {-0.161978191,  -0.193611001},
          {-0.102489983,  -0.114215349},
          {-0.01469968, -0.014918985},
          {-0.008863365,  -0.008942626},
          {0.026751057, 0.026054094},
          {0.166922436, 0.14309643},
          {0.176953808, 0.1504144},
          {0.190233418, 0.159916534},
          {0.199800708, 0.166635331},
          {0.261502822, 0.207600348},
          {0.30182879,  0.232370249},
          {0.83763905,  0.468046718}
    };
    private static readonly Dataset defaultDataset;
    private static readonly IEnumerable<string> defaultAllowedInputVariables;
    private static readonly string defaultTargetVariable;

    private static readonly RegressionProblemData emptyProblemData;
    public static RegressionProblemData EmptyProblemData {
      get { return emptyProblemData; }
    }

    static RegressionProblemData() {
      defaultDataset = new Dataset(new string[] { "y", "x" }, kozaF1);
      defaultDataset.Name = "Fourth-order Polynomial Function Benchmark Dataset";
      defaultDataset.Description = "f(x) = x^4 + x^3 + x^2 + x^1";
      defaultAllowedInputVariables = new List<string>() { "x" };
      defaultTargetVariable = "y";

      var problemData = new RegressionProblemData();
      problemData.Parameters.Clear();
      problemData.Name = "Empty Regression ProblemData";
      problemData.Description = "This ProblemData acts as place holder before the correct problem data is loaded.";
      problemData.isEmpty = true;

      problemData.Parameters.Add(new FixedValueParameter<Dataset>(DatasetParameterName, "", new Dataset()));
      problemData.Parameters.Add(new FixedValueParameter<ReadOnlyCheckedItemList<StringValue>>(InputVariablesParameterName, ""));
      problemData.Parameters.Add(new FixedValueParameter<IntRange>(TrainingPartitionParameterName, "", (IntRange)new IntRange(0, 0).AsReadOnly()));
      problemData.Parameters.Add(new FixedValueParameter<IntRange>(TestPartitionParameterName, "", (IntRange)new IntRange(0, 0).AsReadOnly()));
      problemData.Parameters.Add(new ConstrainedValueParameter<StringValue>(TargetVariableParameterName, new ItemSet<StringValue>()));
      emptyProblemData = problemData;
    }
    #endregion

    public IConstrainedValueParameter<StringValue> TargetVariableParameter {
      get { return (IConstrainedValueParameter<StringValue>)Parameters[TargetVariableParameterName]; }
    }
    public string TargetVariable {
      get { return TargetVariableParameter.Value.Value; }
      set {
        if (value == null) throw new ArgumentNullException("targetVariable", "The provided value for the targetVariable is null.");
        if (value == TargetVariable) return;

        var matchingParameterValue = TargetVariableParameter.ValidValues.FirstOrDefault(v => v.Value == value);
        if (matchingParameterValue == null) throw new ArgumentException("The provided value is not valid as the targetVariable.", "targetVariable");
        TargetVariableParameter.Value = matchingParameterValue;
      }
    }

    public IFixedValueParameter<BoolValue> ScaleInputsParameter {
      get { return (IFixedValueParameter<BoolValue>)Parameters[ScaleInputsParameterName]; }
    }

    public bool ScaleInputs {
      get { return ScaleInputsParameter.Value.Value; }
      set { ScaleInputsParameter.Value.Value = value; }
    }

    public IEnumerable<double> TargetVariableValues {
      get { return Dataset.GetDoubleValues(TargetVariable); }
    }
    public IEnumerable<double> TargetVariableTrainingValues {
      get { return Dataset.GetDoubleValues(TargetVariable, TrainingIndices); }
    }
    public IEnumerable<double> TargetVariableTestValues {
      get { return Dataset.GetDoubleValues(TargetVariable, TestIndices); }
    }

    [StorableConstructor]
    protected RegressionProblemData(StorableConstructorFlag _) : base(_) { }
    [StorableHook(HookType.AfterDeserialization)]
    private void AfterDeserialization() {
      if (!Parameters.ContainsKey(ScaleInputsParameterName)) {
        Parameters.Add(new FixedValueParameter<BoolValue>(ScaleInputsParameterName, "If enabled input features are scaled by a standard transformation (µ=0, σ=1)", new BoolValue(false)));
      }
      RegisterParameterEvents();
    }

    protected RegressionProblemData(RegressionProblemData original, Cloner cloner)
      : base(original, cloner) {
      RegisterParameterEvents();
    }
    public override IDeepCloneable Clone(Cloner cloner) {
      if (this == emptyProblemData) return emptyProblemData;
      return new RegressionProblemData(this, cloner);
    }

    public RegressionProblemData()
      : this(defaultDataset, defaultAllowedInputVariables, defaultTargetVariable) {
    }
    public RegressionProblemData(IRegressionProblemData regressionProblemData)
      : this(regressionProblemData.Dataset, regressionProblemData.AllowedInputVariables, regressionProblemData.TargetVariable) {
      TrainingPartition.Start = regressionProblemData.TrainingPartition.Start;
      TrainingPartition.End = regressionProblemData.TrainingPartition.End;
      TestPartition.Start = regressionProblemData.TestPartition.Start;
      TestPartition.End = regressionProblemData.TestPartition.End;
    }

    public RegressionProblemData(IDataset dataset, IEnumerable<string> allowedInputVariables, string targetVariable, IEnumerable<ITransformation> transformations = null)
      : base(dataset, allowedInputVariables, transformations ?? Enumerable.Empty<ITransformation>()) {
      var variables = InputVariables.Select(x => x.AsReadOnly()).ToList();
      Parameters.Add(new ConstrainedValueParameter<StringValue>(TargetVariableParameterName, new ItemSet<StringValue>(variables), variables.Where(x => x.Value == targetVariable).First()));
      Parameters.Add(new FixedValueParameter<BoolValue>(ScaleInputsParameterName, "If enabled input features are scaled by a standard transformation (µ=0, σ=1)", new BoolValue(false)));
      RegisterParameterEvents();
    }

    private void RegisterParameterEvents() {
      TargetVariableParameter.ValueChanged += new EventHandler(TargetVariableParameter_ValueChanged);
      ScaleInputsParameter.Value.ValueChanged += new EventHandler(ScaleInputsParameter_ValueChanged);
    }
    private void TargetVariableParameter_ValueChanged(object sender, EventArgs e) {
      OnChanged();
    }

    private void ScaleInputsParameter_ValueChanged(object sender, EventArgs e) {
      var transformations = new ShiftStandardDistributionTransformation(Dataset.DoubleVariables);
      transformations.Mean = 0;
      transformations.StandardDeviation = 1;
      var scaling = Transformation.CreateTransformations(transformations, Dataset, TrainingIndices, AllowedInputVariables);

      var scaledVariables = AllowedInputVariables.Select((var, i) => new { Variable = var, Data = scaling[i].Apply(Dataset.GetDoubleValues(var)).ToArray() });
      var newDataset = ((Dataset)Dataset).ToModifiable();

      foreach (var v in scaledVariables) {
        newDataset.ReplaceVariable(v.Variable, v.Data);
      }

      if (!Parameters.ContainsKey("Scaled Dataset"))
        Parameters.Add(new FixedValueParameter<Dataset>("Scaled Dataset", newDataset.ToDataset()));
    }
  }
}